RATES OF FOREST REGROWTH IN EASTERN AMAZÔNIA: A COMPARISON OF ALTAMIRA AND BRAGANTINA REGIONS, PARÁ STATE, BRAZIL JOANNA M. TUCKER, EDUARDO S. BRONDIZIO, and EMILIO F. MORÁN
xtensive areas of secondary vegetation cover the Amazonian landscape and are becoming a prominent feature surrounding rural communities in the Amazon. Due to the shrinking domain of mature forest in colonized areas, farmers today commonly cut secondary forest (fallowed land) rather than mature forest as part of the cycle of shifting cultivation. Secondary forests are an integral part of Amazonian agricultural strategies, not only providing nutrient rich ash when burned, but also facilitating the area’s restoration following abandonment. As a fallow matures organic matter and nutrients accumulate, leaching and erosion are checked, and water and nutrients from lower soil depths are drawn upwards (Smith, 1982; Nepstad et al., 1991). This paper examines rates of forest regrowth, or secondary succession, as a process
which must be understood as a product of both land use and initial environmental endowments. Variance in the speed of forest regrowth is evident across regions and along a soil fertility gradient. The rate of forest succession is determined by several factors. Original floristic composition, neighboring vegetation, and soil fertility and texture may affect regrowth. In addition, farmers’ land use decisions, such as clearing size, clearing procedures, crops planted, frequency of use, and duration of use, influence tree establishment and direct the path of secondary succession. At the regional scale addressed in this paper, soil fertility and land use history emerge as the critical factors influencing the rate of forest regrowth. In this paper we compare and contrast two study sites, Altamira and Bragantina, and examine the relationship between land use in-
tensity and soil fertility and the rate of succession. Differences in tree density, height, and basal area served to distinguish between the rate of succession in Altamira and Bragantina. On the relatively nutrient-rich Alfisols in Altamira, we found a rate of forest regrowth nearly twice as fast as that of the nutrient-poor Oxisols and Spodosols in Bragantina. The two distinct rates put to rest attempts to formulate a uniform scenario for succession in Amazonia, even when confined to the eastern Amazon. However, similarities in structural development during forest regrowth can be found and facilitate understanding of Amazon-wide successional processes. In the literature, it is common to separate successional stages into age classes. Age categories are useful in comparing rates of succession. However, research in the region has
KEY WORDS / Amazônia / Environmental / Brazil / Forest / Joanna Tucker (BS in Environmental Science, Indiana University, 1996) is a graduate student in Latin American Studies at the University of Arizona, Tucson. She began participation in this research three years ago while still an undergraduate at Indiana University. She has returned to the Amazon each year since then. Address: Center for Latin American Studies, University of Arizona, Tucson, AZ 85721 USA. e-mail:
[email protected]) Eduardo S. Brondizio (PhD in Environmental Affairs, Indiana University, 1996) is a postdoctoral fellow and assistant director at the Anthropological Center for Training and Research on Global Environmental Change (ACT), Indiana University, Bloomington. His work has focused on agricultural agroforestry intensification, land use/cover change, and secondary succession in the Eastern Amazon region (especially in the Amazon estuary), and the application of remote sensing to these issues. Address: ACT, Indiana University, Student Bldg. 331, Bloomington, IN 47405,USA. e-mail: ebrondiz@indiana,edu Emilio F. Morán (PhD in Anthropology, University of Florida, 1975) is Rudy Professor of Anthropology, Director of the Anthropological Center for Training and Research on Global Environmental Change (ACT), and Co-director of the Center for the Study of Institutions, Population and Enrivonmental Change (CIPEC) at Indiana University, Bloomington. He has carried out research in Amazônia for the past 25 years. Since 1991 he has incorporated remote sensing to issues of land use and land cover change. Address: (ACT), Indiana University, Student Bldg. 331, Bloomington, IN 47405, USA. e-mail:
[email protected]
64
0378-1844/98/02/064-10 $ 3.00/0
MAR - APR 1998, VOL. 23 Nº 2
shown that age alone cannot be used to predict stages of development of secondary succession, since land use history can strongly influence structural differences among sites within the same age class (Uhl et al, 1988; Brondizio, 1996). Original vegetation, neighboring vegetation, and soil characteristics may also impact fallow development. Physiognomy provides a more consistent way of comparing secondary vegetation across sites and regions. From the point of view of remote sensing and land use analysis, the definition of structural parameters for regrowth stages is especially important, since structural features are associated with spectral data from satellite images. This paper pursues two main goals: 1) to highlight the differences in rates of regrowth between Altamira and Bragantina, and 2) to propose a classification of regrowth stages based on structural criteria that takes into account regional differences in soil fertility and local differences in land use. The paper begins with a characterization of the study areas and is followed by a Methods section. In the Results and Discussion section, we investigate the pattern of secondary growth over time, and explore the variation in the rate of secondary regrowth between Altamira and Bragantina. Next, we present a new cross-regional classification key which divides stages of secondary succession according to physiognomic parameters follows. A discussion of the implications of our findings concludes the paper.
day plant mixed annual crops for one to two years, and afterwards they convert the cropped land to pasture or perennial cash crops (Fearnside, 1986). The second study area surrounds the town of Igarapé-Açu in the micro-political region known as Zona Bragantina within the state of Pará (See Figure 1). It sits between 0°45’S and 1°39’S latitude and 46°16’W and 48°15’W longitude. Rainfall in Bragantina ranges from 2200 to 2800 mm annually. The mean annual temperature of the region is 25°C. According to Köppen’s climate classification system (1936), the Bragantina Region is predominantly type Ami2 (Denich, 1991). Igarapé-Açu is dominated by nutrient-poor Oxisols and Spodosols. Oxisols are estimated to represent 46% and Spodosols 3% of the soils in the Amazon Basin (Nicholaides et al., 1983). Between 1870 and 1910 the rubber-boom caused a population explosion in Belém which greatly increased urban food demand. The government responded with a colonization program to develop agriculture in the Bragantina region (Penteado, 1967). Igarapé-Açu was founded in 1897. Of the almost 1 million hectares of dense tropical forest that covered the Bragantina Region at the beginning of the century, less than 2% remained by 1960 (Penteado, 1967). Today the dominant land use is short-fallow swidden cultivation.
Methods Nested Sampling Strategy The present study employs a nested sampling strategy organized by region, site, plot, and sub-plot to collect field data and link it to TM Landsat satellite images. The region sits at the highest level and represents the greater study area which includes all sample sites. What is designated as a site corresponds to the vegetation stand (fallow or mature forest) selected for sampling. Several tiers of information are gathered at each site: land use history, location (with a GPS (Global Positioning System device), vegetation inventory, and soil samples. To characterize the vegetation, plots (10 x 15 m) are distributed at the site in a stratified random fashion, and sub-plots (5 x 2 m) are nested within them. In the majority of cases ten plots and ten sub-plots were inventoried at each site. Plots are designed to inventory trees, whereas sub-plots are used to inventory saplings, seedlings, and herbaceous species. For the purpose of image analysis, each site becomes a “training sample”, that is, an area of known identity that is used during supervised classification to identify areas of unknown identity. This procedure is detailed elsewhere (Moran et al., 1994a; Brondizio et al., 1996; Moran et al., 1996a).
Study Areas The Altamira study area lies along the Transamazon highway near 3°12’S latitude and 52°13’W longitude in the Brazilian state of Pará (see Figure 1). Annual rainfall for Altamira is approximately 2000 mm with an average temperature of 26°C. According to Köppen’s climate classification system (1936), Altamira exhibits Awi1 climate (Dantas, 1989). The region presents patches of nutrient-rich Alfisols and less fertile Ultisols. All but four of the sampled sites in this paper represent nutrient-rich Alfisols. Alfisols are estimated to represent less than 8% of the soils in the Amazon Basin (Nicholaides et al., 1983) and are the best soils found in the terra firme. Settlers began to colonize the area in 1971 through a government sponsored colonization program. The region has experienced high rates of deforestation and secondary succession associated with implementation of agropastoral projects (Moran et al. 1994b). The majority of small farmers remaining in Altamira to-
MAR - APR 1998, VOL. 23 Nº 2
Fig. 1. Study Areas: Igarapé-Açu (Bragantina Region) and Altamira (Xingu Region)
65
Study Sites To facilitate sampling, secondary vegetation sites were grouped according to time since abandonment: 0-5 years, 5-10 years, and 11+ years. An adequate number of representative samples were chosen from each age class of secondary succession to achieve a well-represented chronosequence of fallow ages. Landsat TM satellite images were used as a reference to encourage even spatial distribution of the sample sites throughout each region. The farmer’s permission to conduct field research on his property was attained prior to any final site selection. Altamira: Fifteen Altamira sites sampled in 1993 and 1992 are incorporated into the analyses, including two mature upland forest sites, one liana forest and one dense moist forest, representing the two types of mature forest present in the region. The successional sites encompass three sites from the 0-5 year age class, six from the 5-10 year age class, and four from the 11+ year age class. Most of the sites were cleared from primary forest during 1971 or 1972, and a few have been recleared from secondary vegetation. Two advanced succession sites serve as control sites. They were cleared from primary forest and then allowed to recover without further human interference as part of a regrowth experiment conducted by EMBRAPA (Empresa Brasileira de Pesquisa Agropecuaría) (Dantas, 1989). The most commonly encountered crop mixture consists of rice, beans, and corn, and land owners often plant both crop and pasture (see Table I for site summaries). Bragantina Sample Sites: Sixteen sites from Igarapé-Açu sampled during 1994 and 1995 are examined in this paper. Only one mature forest site was sampled due to the scarcity of primary forest in the region. Of the fallowed sites, five belong to the 0-5 year age class, four to the 6-10 year age class, and six to the 11+ year age class. All fallowed sites have been cleared more than once. Most of the sites were planted for only 12 years, and all but two successional sites were previously cultivated with a combination of manioc, beans, and corn or rice. Only one abandoned pasture area was sampled (3 year old fallow) (see Table I for site summaries). Farm-Level Interviews In-depth land use interviews with the land owner, or tenant, were conducted at each sample site. Questions were asked to ascertain when
66
the forest or secondary growth was last cut, clearing procedures, when it was burned, the length of prior fallow periods, cultivation and management techniques (such as fertilizer use), types of crops/pasture grown, yields, the time since the land was abandoned, and other pertinent land use history information. Because of the long history of settlement and cultivation, land use histories from Igarapé-Açu farms are limited to the last two or three cultivation cycles.
ing sub-plots in the older fallows. In the majority of cases, the total area sampled remains constant (100m2), but the new method facilitates easier plot by plot comparison between younger and older sites. Plot sizes for all sites are summarized in Table I. Differing sizes exist in some cases due to the ongoing refinement and modifications during the first four years of the project. Data from all the sites used in this paper can be compared for structural analyses.
Vegetation Inventory Soil Samples Plant individuals with diameter-at-breast-height, or DBH > 10 cm are referred to as trees, those with DBH < 10 cm but > 2 cm DBH or 2 m height as saplings, and any plants under 2 cm DBH are grouped as other (includes seedlings, vines and herbaceous plants). In fallows with trees, ten 10 x 15 meter plots were randomly located along a randomly-oriented transect within the forest stand. Inside each plot, all trees were identified and measured for diameter, stem height (height to the first major branch), and total height. Species identification was done in the field by an experienced field botanist. Plant samples were collected, in the case of uncertainty, and later identified by an herbarium technician using the EMBRAPA herbarium or the Museu Paraense Emílio Goeldi Herbarium both in Belém, Pará, Brazil. Height was estimated upon consensus of two or three observers using a five meter rod as a reference. Inside the plot, a subplot of 5 x 2 meters was randomly placed in which all species < 10 cm DBH were counted and identified, and saplings were measured for diameter and total height. In 1993, sub-plots (5 x 2 meters) were placed in half of the plots (5 sub-plots per site), while 10 sub-plots were sampled per site in 1994 and 1995 (one per 10 x 15 meter plot). In the subplots, if the number of individuals were uncountable (such as in the case of grasses), percent coverage was estimated. In young fallows with few if any trees, five 10 x 2 meter plots were distributed similarly as above, each of which was divided into five 2 x 2 meter sub-plots. In each sub-plot the same procedures were carried out as in the 5 x 2 meter sub-plots mentioned above. For the year 1995 data gathered in Bragantina followed slightly modified procedures for young succession. Rather than sampling younger sites in a pattern of five contiguous 10 x 2 meter plots, we set ten 5 x 2 meter plots along a randomly-oriented transect at random distances from one another, thereby imitat-
At each inventoried site, soil samples were collected with a soil auger at 20 cm intervals down to one meter depth. In Altamira, two subsamples were taken at each site and then combined whenever structure and color were consistent between the samples. In Bragantina, one complete sample was taken per site. Soil samples were delivered to the tropical soils laboratories at Centro de Pesquisa Agropecuária do Trópico Úmido da Empresa Brasileira de Pesquisa Agropecuária (EMBRAPA/ CPATU) in Belém, Pará, Brazil, where chemical and physical soil analyses were completed. Results and Discussion Successional Processes Secondary succession associated with shifting agriculture in Amazonia follows a clear pattern of development. During crop or pasture use, burning and weeding delay succession, but after the field is abandoned, the forest begins to regenerate. Secondary vegetation establishes itself through four main processes: regeneration of remnant individuals, germination from the soil seed bank, sprouting from cut or crushed roots and stems, and seed dispersal and migration from other areas. During the first stage of succession, pioneer species such as lightdemanding herbaceous vegetation, seedlings, and saplings occupy the area and compete for space and resources. Secondary species in tropical lowland forests are characterized by a short life cycle, high growth rate, high reproductive resource allocation, small seed size, and long seed viability3 (Gómez-Pompa and VasquezYanes, 1981). Few if any trees (individuals > 10 cm DBH) are present at this stage. A dramatic difference is evident in the second successional stage (generally after five years) which is characterized by a large presence of young trees but still dominated by saplings in terms of density.
MAR - APR 1998, VOL. 23 Nº 2
Table I SUMMARY OF LAND USE HISTORY AND SAMPLING PROCEDURES FOR STUDY SITES IN ALTAMIRA AND BRAGANTINA, PARÁ, BRAZIL.
The trees attract animal seed dispersers thereby facilitating the introduction of new species. Animal dispersal plays a significant role in tree establishment. According to Howe and Smallwood (1982), at least 50% and often 75% or more of tropical forest tree species produce fleshy fruits consumed by birds or mammals. A closing canopy thins the shade-intolerant understory and alters the microclimate by reducing soil temperature and increasing soil moisture, thereby improving conditions for seed germination of mature forest species. Furthermore, gradual restoration of nutrient cycling and accumulation of organic matter facilitate the next stage of development. This third successional stage exhibits a clear shift in structural design. Pioneer tree species die off and are replaced by slow-growing, shade-tolerant mature forest species which favor increased shade
MAR - APR 1998, VOL. 23 Nº 2
and improved edaphic conditions. Trees now dominate the stand, while saplings play a secondary role. The understory consists of saplings and young trees, and fewer individuals are present during this stage. Figure 2 illustrates particular features characterizing the processes of secondary succession in tropical Amazon forests. The advancement of successional growth is significantly impacted by trees which remain throughout the cultivation/pasture cycle. These trees accelerate successional processes through attracting dispersers, especially birds and bats who perch and defecate seeds. Increased seed rain occurs beneath the trees where microclimatic conditions are favorable for germination and growth of late successional and mature forest species. Discussion of the role of remnant trees in
succession, often referred to as tree islands, can be found in Uhl et al. (1982), Uhl (1987), Nepstad et al. (1991), and Vieira et al. (1994). In the two regions discussed in this paper, remnant palms are frequently found in abandoned fields. Common remnant palms include Orbignya phalerata (babaçu) (generally associated with Altamira’s Alfisols) and Maximiliana maripa (inajá) (associated with Oxisols in both regions). Both are well adapted to fire and disturbance initiated by slash-and-burn techniques. Historical and physical conditions of the area are critically important to understanding rates of succession. As shown later in this paper, Altamira forests recover nearly twice as fast as Bragantina forests. In the case of Altamira and Bragantina, regional differences in soil fertility and settlement history are the
67
major factors influencing tree establishment and fallow development. Altamira’s rich Alfisols and recent land use history give this region a successional advantage over the nutrient-poor soils of Bragantina where settlers have exploited the area for 100 years. In Bragantina, mechanisms of colonization, like seed dispersal, seed migration from primary forest, and the seed bank, have broken down under pressure, and tree establishment occurs more slowly as a result. Two Distinct Rates of Secondary Succession: Altamira and Bragantina Regions Density, height, and basal area clearly distinguish between the two rates of secondary forest development in Altamira and Bragantina.
Figure 2. Diagram illustrating key features of secondary succession in Amazônia.
1. Sapling and Tree Density
2. Height
Change in sapling and tree density over time differs between Altamira and Bragantina (see Figure 3). During the age interval between six and 16 years, we found on average, more than 20 saplings per tree in Bragantina, but only half of that in Altamira. Forest regrowth in Altamira exhibits as many as 8600 saplings per hectare by the second year which gradually decline in number as the fallow ages and develops. Bragantina, on the other hand, has at most 6000 saplings per hectare by the third year, and the number of saplings continues to climb until about the tenth year when shade reduces sapling density. The absence of mature forest in Bragantina limits the seed source and may explain slower colonization of woody plants. Denich’s (1991) study of young secondary succession in Bragantina found that only 30 of the seedlings which germinated from seeds within a 10 m2 area were tree, shrub, or liana species. The remaining majority were herbaceous plants. Most secondary woody regrowth in this region result from sprouting. Denich (1991:88) found that all the individuals above 50 cm tall derived from sprouts. The longer time required for sapling establishment in Bragantina, slows the succession process. Despite these differences, a general trend of decline in sapling density over time with a concurrent increase in tree density is evident in both regions. In Altamira there is a remarkably high number of trees by the seventh year of fallow - as many as 773 trees/ha, compared to 407 trees/ha in an eight year old fallow in Bragantina. The larger number of young trees after seven years in Altamira reflects the larger sapling population during the first few years of succession.
Throughout the process of succession, Altamira surpasses Bragantina in height (see Figures 4 and 5). Maximum total height represents the growth potential of a secondary forest. In Altamira, maximum height (10 m) during the first five years of succession is about six meters greater than in Bragantina (3.75 m), while average total height in Altamira (4.19 m) is double that in Bragantina (2.26 m). During the six to ten year period, maximum height is about ten meters higher in Altamira (22.5 m) than Bragantina (13 m). Average total height is the same for a ten year fallow in Altamira and a twenty year fallow in Bragantina (11 m). Thus, Altamira re-
68
growth requires only half the time of Bragantina regrowth to reach the same height. Regional differences are still evident after 15 years of fallow. The average total height of 14-16 year fallows, each cropped for one year and then abandoned, is almost five meters taller in Altamira (13.5 m) than Bragantina (9 m). The maximum height for this age range found in Altamira was about 35 m but only 13 m in Bragantina. Between the two regions average total height is different for the 0-5 year age class (p
